bulk-checkin.c

   1 /*
   2  * Copyright (c) 2011, Google Inc.
   3  */
   4 #include "cache.h"
   5 #include "bulk-checkin.h"
   6 #include "repository.h"
   7 #include "csum-file.h"
   8 #include "pack.h"
   9 #include "strbuf.h"
  10 #include "packfile.h"
  11 #include "object-store.h"
  12
  13 static struct bulk_checkin_state {
  14         unsigned plugged:1;
  15
  16         char *pack_tmp_name;
  17         struct hashfile *f;
  18         off_t offset;
  19         struct pack_idx_option pack_idx_opts;
  20
  21         struct pack_idx_entry **written;
  22         uint32_t alloc_written;
  23         uint32_t nr_written;
  24 } state;
  25
  26 static void finish_tmp_packfile(struct strbuf *basename,
  27                                 const char *pack_tmp_name,
  28                                 struct pack_idx_entry **written_list,
  29                                 uint32_t nr_written,
  30                                 struct pack_idx_option *pack_idx_opts,
  31                                 unsigned char hash[])
  32 {
  33         char *idx_tmp_name = NULL;
  34
  35         stage_tmp_packfiles(basename, pack_tmp_name, written_list, nr_written,
  36                             pack_idx_opts, hash, &idx_tmp_name);
  37         rename_tmp_packfile_idx(basename, &idx_tmp_name);
  38
  39         free(idx_tmp_name);
  40 }
  41
  42 static void finish_bulk_checkin(struct bulk_checkin_state *state)
  43 {
  44         unsigned char hash[GIT_MAX_RAWSZ];
  45         struct strbuf packname = STRBUF_INIT;
  46         int i;
  47
  48         if (!state->f)
  49                 return;
  50
  51         if (state->nr_written == 0) {
  52                 close(state->f->fd);
  53                 unlink(state->pack_tmp_name);
  54                 goto clear_exit;
  55         } else if (state->nr_written == 1) {
  56                 finalize_hashfile(state->f, hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC | CSUM_CLOSE);
  57         } else {
  58                 int fd = finalize_hashfile(state->f, hash, 0);
  59                 fixup_pack_header_footer(fd, hash, state->pack_tmp_name,
  60                                          state->nr_written, hash,
  61                                          state->offset);
  62                 close(fd);
  63         }
  64
  65         strbuf_addf(&packname, "%s/pack/pack-%s.", get_object_directory(),
  66                     hash_to_hex(hash));
  67         finish_tmp_packfile(&packname, state->pack_tmp_name,
  68                             state->written, state->nr_written,
  69                             &state->pack_idx_opts, hash);
  70         for (i = 0; i < state->nr_written; i++)
  71                 free(state->written[i]);
  72
  73 clear_exit:
  74         free(state->written);
  75         memset(state, 0, sizeof(*state));
  76
  77         strbuf_release(&packname);
  78         /* Make objects we just wrote available to ourselves */
  79         reprepare_packed_git(the_repository);
  80 }
  81
  82 static int already_written(struct bulk_checkin_state *state, struct object_id *oid)
  83 {
  84         int i;
  85
  86         /* The object may already exist in the repository */
  87         if (has_object_file(oid))
  88                 return 1;
  89
  90         /* Might want to keep the list sorted */
  91         for (i = 0; i < state->nr_written; i++)
  92                 if (oideq(&state->written[i]->oid, oid))
  93                         return 1;
  94
  95         /* This is a new object we need to keep */
  96         return 0;
  97 }
  98
  99 /*
 100  * Read the contents from fd for size bytes, streaming it to the
 101  * packfile in state while updating the hash in ctx. Signal a failure
 102  * by returning a negative value when the resulting pack would exceed
 103  * the pack size limit and this is not the first object in the pack,
 104  * so that the caller can discard what we wrote from the current pack
 105  * by truncating it and opening a new one. The caller will then call
 106  * us again after rewinding the input fd.
 107  *
 108  * The already_hashed_to pointer is kept untouched by the caller to
 109  * make sure we do not hash the same byte when we are called
 110  * again. This way, the caller does not have to checkpoint its hash
 111  * status before calling us just in case we ask it to call us again
 112  * with a new pack.
 113  */
 114 static int stream_to_pack(struct bulk_checkin_state *state,
 115                           git_hash_ctx *ctx, off_t *already_hashed_to,
 116                           int fd, size_t size, enum object_type type,
 117                           const char *path, unsigned flags)
 118 {
 119         git_zstream s;
 120         unsigned char ibuf[16384];
 121         unsigned char obuf[16384];
 122         unsigned hdrlen;
 123         int status = Z_OK;
 124         int write_object = (flags & HASH_WRITE_OBJECT);
 125         off_t offset = 0;
 126
 127         git_deflate_init(&s, pack_compression_level);
 128
 129         hdrlen = encode_in_pack_object_header(obuf, sizeof(obuf), type, size);
 130         s.next_out = obuf + hdrlen;
 131         s.avail_out = sizeof(obuf) - hdrlen;
 132
 133         while (status != Z_STREAM_END) {
 134                 if (size && !s.avail_in) {
 135                         ssize_t rsize = size < sizeof(ibuf) ? size : sizeof(ibuf);
 136                         ssize_t read_result = read_in_full(fd, ibuf, rsize);
 137                         if (read_result < 0)
 138                                 die_errno("failed to read from '%s'", path);
 139                         if (read_result != rsize)
 140                                 die("failed to read %d bytes from '%s'",
 141                                     (int)rsize, path);
 142                         offset += rsize;
 143                         if (*already_hashed_to < offset) {
 144                                 size_t hsize = offset - *already_hashed_to;
 145                                 if (rsize < hsize)
 146                                         hsize = rsize;
 147                                 if (hsize)
 148                                         the_hash_algo->update_fn(ctx, ibuf, hsize);
 149                                 *already_hashed_to = offset;
 150                         }
 151                         s.next_in = ibuf;
 152                         s.avail_in = rsize;
 153                         size -= rsize;
 154                 }
 155
 156                 status = git_deflate(&s, size ? 0 : Z_FINISH);
 157
 158                 if (!s.avail_out || status == Z_STREAM_END) {
 159                         if (write_object) {
 160                                 size_t written = s.next_out - obuf;
 161
 162                                 /* would we bust the size limit? */
 163                                 if (state->nr_written &&
 164                                     pack_size_limit_cfg &&
 165                                     pack_size_limit_cfg < state->offset + written) {
 166                                         git_deflate_abort(&s);
 167                                         return -1;
 168                                 }
 169
 170                                 hashwrite(state->f, obuf, written);
 171                                 state->offset += written;
 172                         }
 173                         s.next_out = obuf;
 174                         s.avail_out = sizeof(obuf);
 175                 }
 176
 177                 switch (status) {
 178                 case Z_OK:
 179                 case Z_BUF_ERROR:
 180                 case Z_STREAM_END:
 181                         continue;
 182                 default:
 183                         die("unexpected deflate failure: %d", status);
 184                 }
 185         }
 186         git_deflate_end(&s);
 187         return 0;
 188 }
 189
 190 /* Lazily create backing packfile for the state */
 191 static void prepare_to_stream(struct bulk_checkin_state *state,
 192                               unsigned flags)
 193 {
 194         if (!(flags & HASH_WRITE_OBJECT) || state->f)
 195                 return;
 196
 197         state->f = create_tmp_packfile(&state->pack_tmp_name);
 198         reset_pack_idx_option(&state->pack_idx_opts);
 199
 200         /* Pretend we are going to write only one object */
 201         state->offset = write_pack_header(state->f, 1);
 202         if (!state->offset)
 203                 die_errno("unable to write pack header");
 204 }
 205
 206 static int deflate_to_pack(struct bulk_checkin_state *state,
 207                            struct object_id *result_oid,
 208                            int fd, size_t size,
 209                            enum object_type type, const char *path,
 210                            unsigned flags)
 211 {
 212         off_t seekback, already_hashed_to;
 213         git_hash_ctx ctx;
 214         unsigned char obuf[16384];
 215         unsigned header_len;
 216         struct hashfile_checkpoint checkpoint = {0};
 217         struct pack_idx_entry *idx = NULL;
 218
 219         seekback = lseek(fd, 0, SEEK_CUR);
 220         if (seekback == (off_t) -1)
 221                 return error("cannot find the current offset");
 222
 223         header_len = xsnprintf((char *)obuf, sizeof(obuf), "%s %" PRIuMAX,
 224                                type_name(type), (uintmax_t)size) + 1;
 225         the_hash_algo->init_fn(&ctx);
 226         the_hash_algo->update_fn(&ctx, obuf, header_len);
 227
 228         /* Note: idx is non-NULL when we are writing */
 229         if ((flags & HASH_WRITE_OBJECT) != 0)
 230                 CALLOC_ARRAY(idx, 1);
 231
 232         already_hashed_to = 0;
 233
 234         while (1) {
 235                 prepare_to_stream(state, flags);
 236                 if (idx) {
 237                         hashfile_checkpoint(state->f, &checkpoint);
 238                         idx->offset = state->offset;
 239                         crc32_begin(state->f);
 240                 }
 241                 if (!stream_to_pack(state, &ctx, &already_hashed_to,
 242                                     fd, size, type, path, flags))
 243                         break;
 244                 /*
 245                  * Writing this object to the current pack will make
 246                  * it too big; we need to truncate it, start a new
 247                  * pack, and write into it.
 248                  */
 249                 if (!idx)
 250                         BUG("should not happen");
 251                 hashfile_truncate(state->f, &checkpoint);
 252                 state->offset = checkpoint.offset;
 253                 finish_bulk_checkin(state);
 254                 if (lseek(fd, seekback, SEEK_SET) == (off_t) -1)
 255                         return error("cannot seek back");
 256         }
 257         the_hash_algo->final_oid_fn(result_oid, &ctx);
 258         if (!idx)
 259                 return 0;
 260
 261         idx->crc32 = crc32_end(state->f);
 262         if (already_written(state, result_oid)) {
 263                 hashfile_truncate(state->f, &checkpoint);
 264                 state->offset = checkpoint.offset;
 265                 free(idx);
 266         } else {
 267                 oidcpy(&idx->oid, result_oid);
 268                 ALLOC_GROW(state->written,
 269                            state->nr_written + 1,
 270                            state->alloc_written);
 271                 state->written[state->nr_written++] = idx;
 272         }
 273         return 0;
 274 }
 275
 276 int index_bulk_checkin(struct object_id *oid,
 277                        int fd, size_t size, enum object_type type,
 278                        const char *path, unsigned flags)
 279 {
 280         int status = deflate_to_pack(&state, oid, fd, size, type,
 281                                      path, flags);
 282         if (!state.plugged)
 283                 finish_bulk_checkin(&state);
 284         return status;
 285 }
 286
 287 void plug_bulk_checkin(void)
 288 {
 289         state.plugged = 1;
 290 }
 291
 292 void unplug_bulk_checkin(void)
 293 {
 294         state.plugged = 0;
 295         if (state.f)
 296                 finish_bulk_checkin(&state);
 297 }